Episode #34 of 3 MINUTES BULLSHIT WITH GEORGE featured Alper Taner, App Growth Consultant. We discussed how Artificial Intelligence(AI) can help in App Store Optimisation(ASO) and User Acquisition(UA).
In this episode of 3 MINUTES BULLSHIT WITH GEORGE, George Natsvlishvili is joined by Alper Taner, App Growth Consultant.
Alper shares his experience as a Mobile Growth Expert about the AI usage in ASO and UA.
Check out all the other episodes of 3 MINUTES BULLSHIT WITH GEORGE on our website or on Youtube channel.
Connect with Alper here:
Timestamps:
00:00 Dance
00:15 Alper’s Introduction
00:27 AI to Boost ASO & UA
03:39 Bloopers
Transcript & Key takeaways from the episode:
AI tools contribute significantly to idea enhancement and creative direction but do not replace human involvement in the entire process.
For boosting ASO and UA via AI it is important of giving precise system prompts and user prompts to AI as well as a clear definition of input, context, tonality, and output specifications for optimal outcomes.
You shall provide guidelines for text assets including brand guidelines, preferred and disliked examples, character limits, and a thorough definition of output expectations.
Similar principles apply to visuals; providing example images and specifying the balance between generated and user-provided visuals ensures satisfactory results.
For UA Optimization with AI you shall share best and worst performance data, including metrics like CPA, IPM, CTR, and conversion rates as well as provide comprehensive metadata of the app, app description, and contextualize with competition metadata, pinpoint preferences and dislikes in competition's text quality and style.
Then you shall continue iteration and data input to refine AI's output.
Tools Used:
ChatGPT, Writer, Copy.ai, and Perplexity for textual aspects.
Visual tools - DALL·E and Midjourney.
Localization - Heygen for language changes.
George: How do you use AI for ASO and UA?
Alper: Sure. I think it's really important to give the right system prompts and the user prompts to AI and make sure that it understands its role, it understands the input and the context, and the tonality and the output specifications. So the more you define it better outcome you get and it works the same both for the visuals and as well for the text assets. And in the text assets, it is always good to provide the brand guidelines, the examples that you like, the examples that you don't like, and basically everything is well defined, you give the character limits and basically everything that how it should be at the output. And you also make sure that double-check before it sends to you and doesn't make errors and it must be similar to this and it shouldn't do that and all that.
And then when it comes to visuals, it's pretty much similar, like many people complain that “oh the output of Midjourney is not what I expected”, but then you can also feed Midjourney with the example images and tell how much it should use that one versus how much it should generate on. So as long as you provide all these details the outcome is quite satisfactory.
And yeah, so it works both on the ASO and also on the UA pretty much the same way. On the UA side, you provide your best performance, you provide your bad performance and you can also explain what is good and what is bad performer. It might be not just about a CPA, but also the IPM and the CTR and the conversion rates and so on. And of course, you would provide your whole metadata of your like app description and pretty much all the data you have about the app as well as you could give the competition metadata as well to give them a really good context. And then you can also pinpoint what you like about the competition and what you don't like in terms of the text quality and text style. Then you'll be happy and if you're not happy, you can always iterate and you can always give it more data. this movement in terms of men.
George: Cool, man. In terms of tools: what kind of tools you use.
Alper: I would say mostly ChatGPT, but there's a lot out there like Writer, Copy.ai, Perplexity and all that. And then on the visual side, it's mostly the DALL·E, Midjourney. And then for the like for localization, we use a lot of Heygen when we change the language, for example now I speak in English and entonces puedo cambiarlo fácilmente espanol, wir machen es mit AI möglich.
George: Cool, man. So AI tools make your life easier, no?
Alper: Yeah.
George: Both with iterate and with even the visual development.
Alper: Exactly. It's not gonna do everything for you, but it will help to enhance your ideas and it will help you and enhance what you already have, you know, hopefully in a better direction and better like more creative way, but it's not going to do your job for you 100%.
If you have any questions related to online marketing leave your comments/questions below. I will find the expert to answer your question.
Thank you – and feel free to check the other episodes.
Ciao!